Research On Author Information Extraction Based On Deep Belief Nets

With the vigorous development of the book publishing and selling industry, how to choose the high quality books from the massive books has become an urgent problem to be solved. At present, there are many methods to evaluate the book evaluation system, but usually only use a single dimension to evaluate the book. In During many books evaluation model the "five library evaluation system covers a more comprehensive evaluation of the factors on the books. The in the "five dimensional Book Evaluation System" in "the natural information dimension based on further dimensions of book author were in-depth study, construct and book author evaluation related influence factor sequence set. And by using the depth confidence network algorithm, the factor of the sequence is extracted, and the structured book author information is obtained, which provides the technical support for the evaluation of the book author.The research work includes the following aspects:(1) the analysis of Dangdang and Amazon Chinese web, interactive publishing network and the structure of the Chinese teaching material web pages, using the crawler to crawl the book author information as the analysis of the corpus;(2) integrated five dimensional model in book author dimension, excellent book in author information and survey three aspects, summarizes and book author related influence factor;(3) apply deep belief net algorithm to the field of author's information extraction. Based on the influencing factors of the previous paper, the relevant information is extracted from the book author's information. Algorithm is applied in the book the author selected model on the test corpus extraction process, because part of the corpus contains the information is not comprehensive, need to use search engines, using additional information has been extracted entity filled the relevant book author.